Daily AI Security Intelligence

Critical flaws raise credential and AI data-exposure risks

SAP Cloud SDK for Python versions before 0.45.2 contain a CVSS 10 OAuth2 SSRF flaw that can redirect token requests and expose client credentials when tenant input is attacker-controlled, according to the source item “SAP Discloses Critical CVSS 10 Flaw in Cloud SDK for Python Allowing OAuth2 Credential Theft” and corroborating advisories.[1][3] GitLab also patched a CVSS 10 CI/CD regular-expression flaw that could enable unauthenticated server code execution, potentially exposing development pipelines and downstream artifacts.[2][4] The source items further report critical issues in SonicWall SMA 1000, Paperclip, and Cursor that could expose connected environments, privileged agent functions, or developer data, although successful compromise is not established for every case. RealGround analysis: prioritize inventory, patching, secret rotation, access-control validation, and review of untrusted inputs entering AI tools and integrations. These actions reduce the likelihood that credentials, customer data, or AI telemetry are exposed through compromised infrastructure or agent workflows.

2026-10-18 data leakage RealGround analysis
Run AI Security Assessment Talk to AI CISO Download Daily Brief
Top risk today data leakage
Affected industries Healthcare, Fintech, SaaS, SMB, AI startups
Highest severity signal Critical flaws raise credential and AI data-exposure risks
Recommended action Review agent permissions, data access, approval gates, and prompt-injection test coverage.
Relevant RealGround service AI Security Readiness Assessment

What Happened

SAP Cloud SDK for Python versions before 0.45.2 contain a CVSS 10 OAuth2 SSRF flaw that can redirect token requests and expose client credentials when tenant input is attacker-controlled, according to the source item “SAP Discloses Critical CVSS 10 Flaw in Cloud SDK for Python Allowing OAuth2 Credential Theft” and corroborating advisories.[1][3] GitLab also patched a CVSS 10 CI/CD regular-expression flaw that could enable unauthenticated server code execution, potentially exposing development pipelines and downstream artifacts.[2][4] The source items further report critical issues in SonicWall SMA 1000, Paperclip, and Cursor that could expose connected environments, privileged agent functions, or developer data, although successful compromise is not established for every case. RealGround analysis: prioritize inventory, patching, secret rotation, access-control validation, and review of untrusted inputs entering AI tools and integrations. These actions reduce the likelihood that credentials, customer data, or AI telemetry are exposed through compromised infrastructure or agent workflows.

Why This Matters

AI systems increasingly connect natural-language decisions to SaaS integrations, internal data, memory stores, API calls, and production workflows. A signal that appears narrow in a vendor report can become broader business risk when it intersects with autonomous tools or sensitive context.

Healthcare Fintech SaaS SMB AI startups

RealGround Analysis

This trend increases exposure to indirect prompt injection, unauthorized tool execution, sensitive data disclosure, and weak human approval workflows for organizations deploying LLM agents or AI-enabled automation.

Recommended Actions

  • Inventory SAP Cloud SDK for Python usage, upgrade affected deployments to version 0.45.2 or later, validate tenant identifiers against an allowlist, and rotate potentially exposed OAuth2 credentials.
  • Patch self-managed GitLab CE/EE to fixed versions, including 19.4.2, 19.3.5, or 19.2.8 as applicable, then review CI/CD configurations and pipeline artifacts for unauthorized changes.
  • Classify sensitive data before it enters prompts, embeddings, or logs.
  • Apply redaction and retention controls to model inputs and outputs.
  • Use least-privilege access to vector stores and connected databases.
  • Monitor for secrets, customer data, and regulated data in AI telemetry.
  • Restrict agent permissions with least-privilege tool scopes.
  • Add human approval workflows for state-changing actions.
  • Review SaaS integrations, memory persistence, and data access paths.
  • Test prompt injection and indirect prompt injection scenarios before production rollout.

Relevant RealGround Service

AI Security Readiness Assessment Schedule security review AI Agent Business Logic Audit Schedule security review AI Red Teaming Schedule security review AI Agent SBOM and LLM Mapping Schedule security review AI CISO Advisory Schedule security review

Sources

SAP Discloses Critical CVSS 10 Flaw in Cloud SDK for Python Allowing OAuth2 Credential Theft GitLab Patches CVSS 10.0 Flaw That Could Let Unauthenticated Users Run Server Code Critical SonicWall SMA 1000 Flaw Targeted in 135 Exploitation Attempts After Patch Critical Paperclip Flaw Allowed Admin Access, Code Execution Critical Cursor Flaws Could Let Prompt Injection Escape Sandbox and Run Commands
Talk to AI CISO